back to

To get more information contact me at:

Amplifier End version 2  started in the 2021


After the good experience made using the Excicon mosfet ECW20N20 on DC Circlotron I immediately thought about their use in this project.
Using ECW20N20 is possible to reduce the output impedance for a good damping factor and we have lower distortion and more output power.
The DC Circlotron is very complex and I found low interest from DIY but I can understand them.
In any case this new Amplifier End sound much better than DC Circlotron and it is more simple.

This new version of the Amplifier End uses easy-to-find pre-assembled modules for the protection, soft start and filaments power supply.
It was also possible to reduce weight and costs with some changes on power supply section but always without output noise.
In the driver section you can decide to use the Siemens D3a (expensive pentode to use in triode connection) or the Russian 6Z51P / 6Ж51П (cheap pentode tested on Bartola article).
I have used these russian pentode also in the my last Power Folllower and in the Inpol/Mofo and these vacuum tubes sounded better than the NOS Mullard and Siemens. 
Like all my amplifiers also this can be driven directly by a CD, DAC or Phono pre-ampl.

The main benefit of a Circlotron design is an high power amplifier with no switching output stage like a pure class A but higher efficiency, with only 1A bias current still 1mA with 2.3A on 8 load.


Pout = 60w on 8ohm with 1.5% thd
Pout = 105W on 4ohm with 1.6% thd
Pout = 21W on 8ohm no-switching with 1A bias
Pd = output stage 2 x 1A x 38V = 76W each channel with 1A bias

Zout = 0.4ohm to have a damping factor of 20.
Ft(-3dB) = 20Hz-56KHz at 0dB and  10Hz-65KHz at -3dB


Please do not touch with multimeter probe the mosfet gate after the power on because this operation will destroy the mosfets.
Before to solder the mosfets set the trimmer to have 0.2VDC on all trimmer outputs,  then turn off the amp., unload the capacitors of the power stage, solder the mosfets, screw them on the heatsink with a silicon pad isolator and after the power on increase both trimmers, any time one turn for each, until you get the correct bias and turn only one trimmer to get zero DC voltage on output terminals (after check again the current to get 1A).


These companies have supported this project.




It is very simple, only few components on the signal path.

The interstage transformer is used to reduce the output impedance of the vacuum tube and to create a differential signal to drive the mosfet with opposite phase as required by the Circlotron design.

D3a and 6Z51 tubes in the triode connection have about 70 - 80x of voltage gain and an internal resistance about 2 - 2.8Kohm.




For this design is necessary a tube with high amplification factor > 60x and low internal resistance < 3000ohm.

The socket pins are not the same for both the tubes.


 I suggest to buy a large stock of 6Z51P to make a selection on voltage gain.



The interstage transformer is used to reduce the output impedance of the vacuum tube and to create a differential signal to drive the mosfet with opposite phase as required by the Circlotron design.

It is necessary a good 4 : 1+1 with a minimun inductance of 50H to have a good low frequency response.

Ft(-3dB) = R / ( 2 * 3.14 * L ) = 3000ohm / ( 2 * 3.14 * 50 ) = 9.55Hz

LL1671/30mA in ALT Q configuration have 35H so the version with an air gap for only 20mA will have about 35H * 30mA / 20mA = 52.5H.

Using a 4 : 1 instead of a 2 : 1 + 1 the mosfet grid will be driven with low impedance and more current so there is a better high frequency response and faster slew rate.



The mosfet used in the previous release was the 2SK1058 Hitachi Renesas.
This new version use the ECW20N20 Exicon with lower output impedance and the same dynamic range of the 2SK1058.

Simulation of the output impedance with the old 2SK1058 and with the new Exicon ECW20N20.




Using a CLC power supply in a normal amplifier you can reduce the noise but in this case the mosfet gate is isolated from the high current power supply so there is no benefit and it is possible to use a single 22000uF or 10000uF + 15000uF like in my case.

Following the experince of my last amplifiers I have used Schottky Fast Soft recovery diodes in any section of the power supply but you can use normal 36A diode bridge like the IRF 36MB60 and for the anodic voltage a diode bridge made with 4 x UF5406.

To create a fast diode bridge with these TO220 and large SMD chips I have created 2 simple pcb.

For this design you can use normal transformers produced in stock  but my choice is 2 custom made on my specifications:

Italtras Srl Transformers in the world
30-0       6A
30-0       6A
9-0         2A
115-0     0.3A

external size 125x65mm, core type M0, flux 80% than normal (13000 gauss instead of 17000 gauss).

There are some choice for the power supply capacitors:






All the resistors on the signal path have 2 positions on schematic because ...

otherwise to get a slightly better result using the MK132 Caddock.




The my choice is the HiFi-2000 mod. Pesante dissipante 5 units with 10mmm frontal panel.
On paper each lateral heat sink of this chassie have a capacity of 0.25 C/W so it will keep the output devices in a good safe area also when the environment condition are terrible 35 C.







Any serious solid state amplifier need a protection circuit because a fault on output transistors or mosfet can destroy the loudspeakers.

The Cirlclotron design never will produce an high dc output but I suggest to use a dc protection circuit.

I have decided to use 2 x AIYIMA 2.0 Digital Power Amplifier Speaker Protection Board Delay Relay Speaker Protection available on Alixpress online shop. 

This module use 2 optoisolator
PC817 for each input and are necessary only some little changes to increase the start-up time.

Attection 9VAC used for the filaments are not enought for this module so use for both 11 or 12VAC.




I am using this module got on Alixpress because it incluse a soft-start and a temperature protection at 75.

Vandal Resistant  Push Button Switches

RS Stock No. 174-6381

Manufacturer Apem Manufacturers

Part No. AV021003C900

The panel cut out  is 22mm, the same of noval socket, so this is the ideal choice for the 10mm front panel



These are the measurement of the distortion and frequency response of the driver stage on one secondary of interstage transformer.


Here we have a voltage gain of 25.6dB and about 8Hz - 60Khz at -3dB (Ck create a little peak at 13Hz).

Here we have a voltage gain of 26.4dB and about 10Hz - 60Khz at -3dB.







The total cost to create this amplifier is not very low but you are building a reference amplifier without compromise.

You can reduce the cost using only 4 x 22000uF 50V KG for the high current power supply, another attenuator and standard transformers.


Qty model brand unit price (euro) total price (euro) shop
2 6Z51 - 5 10
4 ECW20N20 Excicon 12 48 Electronic shop
2 LL1671/20mA Lundahl 100 200 Lundahl distributors
1 10K+10K DACT 145 145  
2 159M Hammond 21 42  
4 10000uF 50V Nichicon Super Through 15 60
4 15000uF 50V Nichicon Super Through 20 80
4 2.2uF 250V MKP Solen 2     6
2 Custon TR 130 260
16 MBR40250 ON 2 32
8 MBRS3200T3 ON 1 8
1 chassie Hi-Fi2000 250 250
2 protection module - 10 20
1 LM317 soft start  - 10 10
1 Soft start   10 10
1 connectors   30 30  
1 other items   50 50  
      total 1261